What is Copilot?
Copilot is our latest AI-powered feature designed to help users quickly interpret study results and make data-driven creative decisions. Part of Studio 3.0, Copilot provides clear, actionable summaries and recommendations tailored to each asset being tested.
Instead of spending time deciphering complex reports, users can now get instant insights and focus on making impactful design choices.
How Accurate is Copilot?
Copilot is built on Claude, a state-of-the-art large language model (LLM) hosted securely on Amazon Bedrock. To ensure the highest accuracy, Copilot's insights are generated using:
Your uploaded assets – Providing context for analysis.
Dragonfly’s predictive data – Industry-leading cognitive prediction algorithms.
Dragonfly’s methodology – Scientifically backed marketing principles.
Industry best practices – Insights from Byron Sharp’s work on creative effectiveness.
Benchmark data – Comparative insights based on historical trends. * coming soon
This combination ensures Copilot delivers reliable, evidence-based recommendations rather than arbitrary suggestions.
How Does Copilot Work?
When a user creates a study in Studio 3.0 and uploads assets, Copilot analyzes:
The uploaded assets to understand the creative context.
Predictive data and study results to assess attention, clarity, and other key metrics.
Marketing and industry principles to provide best-practice recommendations.
The result? A streamlined workflow where users can instantly access executive summaries and data-backed suggestions to enhance their creative assets.
How to Handle Inaccurate or Unexpected Insights from Copilot
While Copilot is designed for accuracy, AI-generated insights may sometimes require human validation. If a recommendation seems off, consider:
Reviewing the study parameters to ensure the correct context was set.
Cross-referencing the results with your brand guidelines and objectives.
Manually adjusting the AOI (Areas of Interest) to refine results.
Copilot is meant to augment decision-making, not replace human expertise. It provides an excellent starting point, but users should apply their domain knowledge for final decisions.
What if I Have a Suggestion for Copilot Results?
We value user feedback and continuously improve Copilot’s capabilities based on real-world use cases. If you have suggestions, you can:
Share feedback through our [email protected]
Participate in our beta testing programs.
By integrating user input, we ensure Copilot remains a cutting-edge tool that evolves alongside the needs of our customers.